The Same Goods, Landed Twice, at Two Different Costs
Two containers of the same product, from the same supplier, on the same day. They cleared at different ports and they did not cost the same — and the system holding one cost per item has already averaged the difference away.
There is an assumption buried so deep in inventory accounting that almost nobody states it: that the cost of getting goods into a country is a property of the country. Duty is national. The rate is published. Two consignments of the same item, bought at the same price, arrive having paid the same charges — so any difference between them is freight, or timing, or the exchange rate, and those are the things a system is built to track.
In Somalia that assumption does not hold, and it does not fail at the border. It fails inside it.
Berbera, Bosaso and Mogadishu are administered separately. What a consignment cost to bring ashore depends on which of them cleared it. Not on the supplier, not on the incoterm, not on the week — on the port. Every importer here already knows this, because they pay it. The question this post is about is narrower and less obvious: what does your system do with two true costs for one item?
What this post does not contain
Any duty figure, for any of the three. We hold no tariff schedule and we have not read one, and a differential invented to make a blog post concrete would be the fastest possible way to make it wrong. The argument needs only the fact that the two charges differ, and you have that fact on two invoices in your own file. Your forwarder is the authority on the numbers, per consignment, and they will tell you.
What a moving average does with two truths
Most systems hold one cost per item and update it as stock arrives — a weighted average, recalculated on receipt. That mechanism is correct and it is not the problem. The problem is what it is being asked to represent.
A weighted average exists to smooth price variation: the same goods cost a little more this month than last, and you do not want your margin jumping around because of which pallet a picker happened to reach for. It assumes the variation is noise. Here the variation is not noise — it is a structural, repeating, knowable difference with a cause you can name. And an average of two structural costs describes neither of them.
| What you have | What one cost per item gives you | What it costs you |
|---|---|---|
| Consignment A, cleared at one port | Blended into the average on receipt | Its real cost is no longer recoverable from the system — only reconstructible from paperwork |
| Consignment B, cleared at another | Blended into the same average | Same, and now the two are indistinguishable |
| One selling price | Measured against the average | Whichever consignment was the more expensive one has been funding the margin you thought you made on the other |
| A monthly margin report | Plausible. Internally consistent. Wrong in a fixed direction per port | Nothing looks broken, so nobody looks |
The error never announces itself. The average is always plausible, the report always balances, and the only two numbers that are wrong are the two you make decisions with.
The illustrative version, and why the real one is yours
Take an item bought at 100 a unit, 1,000 units per consignment, two consignments. Say the total charges to get consignment A onto your own floor come to 18 a unit and consignment B to 26 — numbers chosen to be arithmetic rather than research. A blends to 118, B to 126, and the system carries 122.
Now price at a 20% margin on cost. Against 122 you sell at 146.40. On the A stock you are making 24%. On the B stock you are making 16%. You believe you are making 20% on both, and no report you run will contradict you, because the report is built from the same 122.
Change the mix and it moves again. A quarter in which more of your volume came through the higher-charge route produces a margin miss with no visible cause — nothing in purchasing changed, nothing in pricing changed, and the variance report has nowhere to put it. This is the specific shape of the problem: a real, repeating cost difference with no column to live in.
Why the usual fixes do not fix it
-
Separate item codes per port
It works, once, and then it spreads. Two codes for one physical product breaks stock enquiry, reorder points, sales reporting and every customer-facing document, and it puts the burden on whoever picks. The cost dimension you needed has been forced into the identity of the goods, which is the wrong place for it.
-
A spreadsheet alongside
The real costs live in a file that is correct and disconnected. It is right until the person maintaining it is on leave, and it never reconciles to the ledger because it was never in it.
-
Book the difference to overheads
Now the cost has left the goods entirely. Gross margin is uniformly overstated on everything, and the number in overheads cannot be attributed to any consignment, port or period. This is the most common answer because it is the easiest, and it is the one that destroys the most information.
-
Average and accept it
A defensible choice if the difference is small. Nobody who has run two clearance routes for a year thinks it is small — and if it is genuinely immaterial, you should be able to say so with a figure rather than an instinct. The point of the test below is to produce that figure.
The test, which needs no vendor and no demonstration
Pick one product you have imported through two different ports. For each consignment, add up everything you paid to get it onto your own floor — supplier invoice, freight, clearing, duty and charges, handling, storage, inland transport. Two numbers.
- Can you produce both figures at all? If not, that is the finding, and it cost you nothing to discover.
- How far apart are they? That gap is your exposure per unit, and multiplying it by the volume that came through the more expensive route gives you the annual figure.
- What price have you been selling that item at, and which of the two costs does it actually clear?
- When the mix shifted last year, did anything in your reporting show you why the margin moved?
- Ask your forwarder to confirm the charge difference is structural rather than a one-off. If it is structural, it needs a column. If it is a one-off, an average is fine and you have just proved it.
That arithmetic is yours and it is worth more than any demonstration, because it either produces a number that changes what you do or it produces the reassurance that this is not your problem. Both are useful and neither requires buying anything.
What software can and cannot do about it
The honest division is narrow, so here it is with the boundary drawn in the same paragraph as the claim.
Cost held per consignment, open after receipt
Freight, duty, clearing, handling and storage attach to the specific consignment as each invoice arrives — often weeks after the goods — and the unit cost for that receipt recalculates. Two landings of one item keep two true costs instead of resolving into one that describes neither.
The paperwork against the consignment
Clearing documents held against the receipt they belong to and retrievable by the consignment rather than by whoever filed them, so the composition of a cost can be re-checked a year later.
The port as a reportable attribute
Comparing margin across clearance routes as a dimension you can filter and group by, rather than by exporting two receipts and doing it by hand. This is a defined build rather than a live feature, and it is commissionable on a written specification, a timeline and a price.
Any duty calculation, lookup or schedule
Not ours and not planned. We record what you were charged. We do not derive it, we hold no tariff table for any administration, and we have no connection to any of them.
Which authority applies, and what you owe
Not ours at any price. There is more than one administration collecting revenue and the arrangements between them are not something we characterise. Your adviser and your forwarder deal with it per consignment, and they are the ones who know.
A vendor question that separates people quickly
Ask any vendor: "if I receive the same item through two ports at two different total costs, what does your system hold?" The useful answers are "one cost, blended, and here is where the difference goes" or "cost per receipt, and here is how you compare them". Both are honest. The answer to be careful with is a confident claim to handle per-port duty automatically — that requires a maintained tariff schedule for three separate administrations, and it is worth asking to see it.
The general form, because it is not really about Somalia
The mistake underneath this is assuming the unit a cost varies over. Most systems assume cost varies by supplier, by date and by currency, because in most markets it does. Somalia varies it by route, and once you have noticed that the unit can be something else, you start finding others: a rate that varies by the customer's registration status, a charge that varies by which state the goods moved between, a levy computed on a base that includes another levy.
The question to carry into any new market is not "what is the rate" but "what does this vary by, and does my system have a place to put that". Somalia is a clean case because the answer is a physical thing you can point at. Most are not that obliging.
What is built, what is not, and what we would decline is set out on the Somalia market page, including the two boundaries above stated as boundaries rather than as roadmap. The mechanics of costs arriving after the goods are in the invoices that arrive three weeks after the goods, and the general method is in what is landed cost.
